US20110142368A1 - Method and apparatus for block-based image denoising - Google Patents

Method and apparatus for block-based image denoising Download PDF

Info

Publication number
US20110142368A1
US20110142368A1 US12913909 US91390910A US2011142368A1 US 20110142368 A1 US20110142368 A1 US 20110142368A1 US 12913909 US12913909 US 12913909 US 91390910 A US91390910 A US 91390910A US 2011142368 A1 US2011142368 A1 US 2011142368A1
Authority
US
Grant status
Application
Patent type
Prior art keywords
weights
reference blocks
current block
block
blocks
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
US12913909
Other versions
US8818126B2 (en )
Inventor
Gun-Shik Shin
Dong-Hyun Kim
Jae-Hun Lee
Sang-yun Kim
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Samsung Electronics Co Ltd
Original Assignee
Samsung Electronics Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00Image enhancement or restoration, e.g. from bit-mapped to bit-mapped creating a similar image
    • G06T5/001Image restoration
    • G06T5/002Denoising; Smoothing
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20021Dividing image into blocks, subimages or windows

Abstract

A block-based image denoising method includes determining similarities between a current block and reference blocks within a search range around the current block, from among certain-sized blocks divided from an input image; determining weights of the reference blocks with respect to the current block based on the similarities; and generating resultant blocks by denoising the current block with respect to every block of the input image based on the weights of the reference blocks.

Description

    CROSS-REFERENCE TO RELATED PATENT APPLICATION
  • This application claims the benefit of Korean Patent Application No. 10-2009-0125701, filed on Dec. 16, 2009, in the Korean Intellectual Property Office, the disclosure of which is incorporated herein in its entirety by reference.
  • BACKGROUND
  • 1. Field
  • The exemplary embodiments relate to an image processing method and apparatus for image and video denoising.
  • 2. Description of the Related Art
  • For image denoising, a local linear minimum mean square error (LLMMSE)-based filter and a bilateral filter are often used. In an image denoising method using an LLMMSE-based filter or a bilateral filter, pixel-based image processing is performed and then an original pixel value is updated to a filtered pixel value.
  • As full high definition (FHD) level 1920×1080 resolution video data becomes increasingly popular and the development of ultra definition (UD) level 3840×2160 resolution video data or higher is considered, high speed is required in image denoising for high-quality image processing and real-time image processing.
  • SUMMARY
  • The exemplary embodiments provide a method and apparatus for block-based image denoising.
  • According to an aspect of the exemplary embodiment, there is provided a block-based image denoising method including determining similarities between a current block and reference blocks within a search range around the current block, from among certain-sized blocks divided from an input image; determining weights of the reference blocks with respect to the current block based on the similarities; and generating resultant blocks by denoising the current block with respect to every block of the input image based on the weights of the reference blocks.
  • The determining of the similarities may include determining a similarity between the current block and each of neighboring blocks disposed within the search range around the current block. The determining of the similarities may include determining a sum of absolute differences (SAD) between the current block and the reference blocks.
  • The determining of the weights may include determining exponential function values with respect to the SAD between the current block and the reference blocks, as the weights of the reference blocks. The changing of the weights may include changing the weights of the reference blocks by performing soft thresholding, before regularization is performed to calculate a weighted average of the reference blocks with respect to the current block.
  • The determining of the weights may include changing the weights of the reference blocks in order to reduce differences in boundaries between the resultant blocks obtained by denoising the current block and neighboring blocks.
  • The generating of the resultant blocks may include generating a resultant block with respect to the current block by using the reference blocks and a weighted average of the weights of the reference blocks.
  • Denoising is performed on each of color channels of the input image.
  • According to another aspect of the exemplary embodiment, there is provided a block-based image denoising apparatus including a similarity determination unit for determining similarities between a current block and reference blocks within a search range around the current block, from among certain-sized blocks divided from an input image; a weight determination unit for determining weights of the reference blocks with respect to the current block based on the similarities; and a resultant block generation unit for generating resultant blocks by denoising the current block with respect to every block of the input image based on the weights of the reference blocks.
  • The weight determination unit may change the weights of the reference blocks in order to reduce differences in boundaries between the resultant blocks obtained by denoising the current block and neighboring blocks.
  • According to another aspect of the exemplary embodiment, there is provided a computer readable recording medium having recorded thereon a computer program for executing the method.
  • BRIEF DESCRIPTION OF THE DRAWINGS
  • The above and other features of the exemplary embodiments will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings in which:
  • FIG. 1 is a block diagram of a block-based image denoising apparatus, according to an exemplary embodiment;
  • FIG. 2 is a diagram for comparing a pixel-based similarity determination method and a block-based similarity determination method;
  • FIG. 3 is a diagram for comparing a pixel-based updating method and a block-based updating method;
  • FIG. 4 is a diagram for describing a weight determination method, according to an exemplary embodiment; and
  • FIG. 5 is a flowchart of a block-based image denoising method, according to an exemplary embodiment.
  • DETAILED DESCRIPTION OF THE EXEMPLARY EMBODIMENTS
  • Hereinafter, the exemplary embodiments will be described in detail with reference to the attached drawings.
  • FIG. 1 is a block diagram of a block-based image denoising apparatus 100, according to an exemplary embodiment.
  • The block-based image denoising apparatus 100 includes a similarity determination unit 110, a weight determination unit 120 and a resultant block generation unit 130. The block-based image denoising apparatus 100 divides an input image into certain-sized blocks and performs block-based image processing.
  • The similarity determination unit 110 determines similarities between a current block and reference blocks from among the blocks of the input image.
  • The current block processed by the similarity determination unit 110 refers to a block that is currently processed from among blocks that are non-overlapping and sequentially processed. A reference block of the current block is one of blocks within a search range around the current block, and the similarity determination unit 110 may determine a similarity between the current block and each of one or more reference blocks within the search range. The reference blocks may be overlappingly selected from within the search range.
  • The search range around the current block may include a temporally or spatially neighboring region of the current block. Also, a region of an original image or a region of an image formed of previously denoised blocks may be determined as the search range so as to be recursively used for image denoising of a next block. Hereinafter, for convenience of explanation, it is assumed that the similarity determination unit 110 determines the similarities between the current block and the reference blocks based on a spatially neighboring region as the search range in the input original image.
  • However, the search range around the current block is not limited to the spatially neighboring region of the current block and may also include a temporally or spatially adjacent region of the current block.
  • The similarity determination unit 110 may use a sum of absolute differences (SAD) between the current block and the reference blocks in order to determine the similarities between the current block and the reference block.
  • The similarities determined by the similarity determination unit 110 are output to the weight determination unit 120.
  • The weight determination unit 120 determines weights of the reference blocks with respect to the current block based on the similarities input from the similarity determination unit 110.
  • The weight determination unit 120 may determine exponential function values of the SAD between the current block and the reference block as the weight of the reference blocks.
  • The weight determination unit 120 may additionally change the weights of the reference blocks in order to reduce differences in boundaries between resultant blocks obtained by denoising the current block, and neighboring blocks. For example, before regularization is performed to calculate a weighted average of the reference blocks with respect to the current block, the weight determination unit 120 may change the weights of the reference blocks by performing soft thresholding. In this case, a threshold value used to perform soft thresholding may be determined in consideration of average luminance of each of the current block and the neighboring blocks.
  • The weights determined by the weight determination unit 120 are output to the resultant block generation unit 130.
  • The resultant block generation unit 130 generates resultant blocks by denoising the current block. The resultant blocks are generated with respect to every block of the input image based on the weights input from the weight determination unit 120. In this case, the resultant blocks processed by the resultant block generation unit 130 correspond on a one-to-one basis with the blocks of the input image and thus are not overlapping in the same manner that the blocks of the input image are not overlapping.
  • The resultant block generation unit 130 may generate and output a resultant block by denoising the current block by using the reference blocks of the current block and a weighted average of the reference blocks.
  • The block-based image denoising apparatus 100 may perform denoising on each of color channels of the input image. For example, in order to denoise an YCbCr image, the block-based image denoising apparatus 100 may perform block-based image denoising on each of Y, Cb and Cr channels. Also, in order to denoise an RGB image, the block-based image denoising apparatus 100 may perform block-based image denoising on each of R, G and B channels.
  • In a resultant image formed of the resultant blocks generated by the resultant block generation unit 130, as block-based image processing is performed, the amount of calculation may be reduced and noise may be effectively removed. Also, as soft thresholding is performed on weights, a blocking effect that occurs in boundaries between the resultant blocks may be effectively reduced.
  • FIG. 2 is a diagram for comparing a pixel-based similarity determination method and a block-based similarity determination method.
  • In the block-based image denoising apparatus 100 illustrated in FIG. 1, the similarity determination unit 110 determines a similarity between a current block and a reference block and the weight determination unit 120 determines a weight of the reference block with respect to the current block. For example, when spatially neighboring image data is used as reference image data, if image processing is performed in units of pixels, a similarity between a current pixel 210 and a neighboring pixel 220 may be measured and then a weight of the neighboring pixel 220 with respect to the current pixel 210 may be determined accordingly.
  • On the other hand, in the block-based image denoising apparatus 100, the similarity determination unit 110 determines a similarity between a current block 215 and a neighboring block 225 and the weight determination unit 120 determines a weight of the neighboring block 225 with respect to the current block 215 based on the similarity.
  • FIG. 3 is a diagram for comparing a pixel-based updating method and a block-based updating method.
  • In the block-based image denoising apparatus 100 illustrated in FIG. 1, the resultant block generation unit 130 outputs resultant blocks obtained by denoising a current block with respect to every block of an input image, in units of blocks. If image processing is performed in units of pixels, a current pixel 310 may be updated to a resultant pixel 320 by denoising the current pixel 310. On the other hand, in the block-based image denoising apparatus 100, the resultant block generation unit 130 updates a current block 315 to a resultant block 325 by denoising the current block 315.
  • As described above in relation to FIGS. 2 and 3, since similarity determination, weight determination, and denoised resultant block generation are performed in units of blocks, in comparison to a pixel-based image denoising apparatus, the block-based image denoising apparatus 100 may accurately detect edge regions and texture regions, and thus may reduce damage to an image signal.
  • Accordingly, in comparison to a pixel-based image denoising apparatus, the block-based image denoising apparatus 100 may realize a visually-natural connection between pixels even when image processing is performed on a region including a plurality of pixels, e.g., an edge region or a texture region.
  • Also, the block-based image denoising apparatus 100 may effectively remove noise that occurs over a plurality of pixels, e.g., film grain noise or scaled noise, by using a block that is larger than the noise. Also, in comparison to a pixel-based image denoising apparatus, the block-based image denoising apparatus 100 may reduce the amount of calculation due to block-based image processing.
  • Meanwhile, in the block-based image denoising apparatus 100, when the weighted average of the reference blocks with respect to the current block is calculated, the resultant block generation unit 130 performs regularization such that a sum of the weights of the reference blocks has a value 1. However, due to differences in image characteristics such as luminance between the current block and the reference blocks, a blocking effect may occur in boundaries between the resultant blocks. This is because, for example, after the weights of the reference blocks are regularized to calculate the weighted average, when blocks having low luminance levels from among original blocks are denoised, variations in luminance before and after denoising is performed are relatively large and thus denoising levels of blocks are different.
  • Accordingly, in order to prevent the blocking effect in the boundaries between the resultant blocks, before regularization is performed to calculate the weighted average, the weight determination unit 120 may adjust the weights of the reference blocks.
  • FIG. 4 is a diagram for describing a weight determination method of the weight determination unit 120 illustrated in FIG. 1, according to an exemplary embodiment.
  • In first through third graphs 400, 480 and 490, a horizontal axis represents the location of a representative pixel of a block and a vertical axis represents a weight. The first through third graphs 400, 480 and 490 show weight with respect to the locations of representative pixels of blocks according to operational steps. In FIG. 4, for convenience of explanation, it is assumed that a spatially neighboring region around a current block is selected as a search range and that one of neighboring blocks of the current block within the search range is selected as a reference block.
  • The first graph 400 shows initial weights 410, 420, 430, 440, 450, 460 and 470 determined by the weight determination unit 120. With respect to blocks that are spatially included within the search range around the current block, the initial weight 410 of the current block and the initial weights 420, 430, 440, 450, 460 and 470 of neighboring blocks may be determined.
  • The weight determination unit 120 may determine the initial weights 410, 420, 430, 440, 450, 460 and 470 as represented by Equation 1.
  • w 1 ( i , m ) = exp ( - y ( i ) - y ( m ) l 1 2 σ n 2 ) [ Equation 1 ]
  • When M is an integer greater than a value 0, within a search range formed of M blocks, y(i) represents an ith block, i.e., a current block, and y(m) represents an mth block, i.e., a neighboring block. Also, i and m are integers greater than a value 0 and equal to or less than M.
  • {right arrow over (y)}(i) represents a vector of the current block y(i) and {right arrow over (y)}(m) represents a vector of the neighboring block y(m).
  • ∥−∥l1 is a function for outputting an SAD of an input and represents a similarity between the current block and the neighboring block.
  • νn 2 an represents a variance value. The weight determination unit 120 may use a fixed variance value or a variance value measured with respect to the current block and the neighboring block.
  • The weight determination unit 120 may determine a weight of the neighboring block with respect to the current block in the form of a Gaussian curve function by using the similarity between the current block and the neighboring block and an exponential function using a variance value as an input. Consequently, the weight determination unit 120 may determine W1(i,m) as an initial weight of the neighboring block, i.e., the mth block with respect to the current block, i.e., the ith block.
  • The second graph 480 shows intermediate weights 412 and 452 adjusted due to soft thresholding performed by the weight determination unit 120. The weight determination unit 120 may obtain the intermediate weights 412 and 452 by performing soft thresholding on the initial weights 410, 420, 430, 440, 450, 460 and 470. In more detail, soft thresholding may be performed as represented by Equation 2.
  • w 2 ( i , m ) = Soft ( w 1 ( i , m ) , th ) = { w 1 ( i , m ) - th , if w 1 ( i , m ) > th 0 , otherwise [ Equation 2 ]
  • Soft( ) is a soft thresholding function and is defined as represented by Equation 2. In more detail, as soft thresholding is performed on the initial weight W1(i,m) by using the function Soft( ) using a threshold value th as an input, W2(i,m) is determined as a value 0 with respect to the initial weights 420, 430, 440, 460 and 470 that are less than the threshold value th, and is determined as the intermediate weights 412 and 452 with respect to the initial weights 410 and 450 that are greater than the threshold value th by subtracting the threshold value th from the initial weights 410 and 450.
  • The third graph 490 shows ultimate weights 415 and 455 obtained by performing regularization. In more detail, as regularization is performed on intermediate weights generated by performing soft thresholding, ultimate weights of neighboring blocks of a current block within a search range may be determined. According to the third graph 490, only the intermediate weights 412 and 452 that do not have a value 0 may be regularized into the ultimate weights 415 and 455 having valid values.
  • In the block-based image denoising apparatus 100, the resultant block generation unit 130 may generate a resultant block by denoising the current block based on the weights determined by the weight determination unit 120 and the weighted average calculated from the weights.
  • The resultant block generation operation of the resultant block generation unit 130 may be performed as represented by Equation 3.
  • x ( i ) = 1 Z m N i w 2 ( i , m ) y ( m ) [ Equation 3 ]
  • Ni represents a search range around a current block and Z is a regularization factor for maintaining a sum of weights as a value 1. Accordingly, a resultant block x(i) obtained by denoising the current block y(i) may be determined based on a weighted average of neighboring blocks y(m) within the search range Ni by using weights W2(i,m) for preventing a blocking effect that occurs in boundaries between resultant blocks x(i).
  • The weight determination unit 120 may optionally adjust the weights to prevent a blocking effect that occurs in boundaries between the resultant blocks. Accordingly, the resultant block generation unit 130 may determine the resultant blocks as represented by Equation 4 by using the weights determined by the weight determination unit 120 without adjusting the weights for preventing the blocking effect.
  • x ( i ) = 1 Z m N i w 1 ( i , m ) y ( m ) [ Equation 4 ]
  • Accordingly, a resultant block x′(i) obtained by denoising a current block without considering a blocking effect that occurs in boundaries between resultant blocks x′(i) may be generated by using initial weights w1(i,m) determined based on similarities between the current block and neighboring blocks y(m) and a weighted average of the neighboring blocks y(m).
  • If noise exists in image data, in an image signal having noise, an average value of the image signal is maintained constant and an average value of the noise is always 0. As such, in order to effectively perform image denoising, statistically, a block size may be equal to or greater than a certain size.
  • Accordingly, the block-based image denoising apparatus 100 is based on certain-sized blocks and determines similarities between a current block and reference blocks by using weights instead of on/off states and image processing using weights may allow non-artificial, i.e., natural, image denoising.
  • The block-based image denoising apparatus 100 is more effective for grain denoising in comparison to a pixel-based image denoising apparatus. Also, the block-based image denoising apparatus 100 regards a region, in which a significant stair step effect occurs, as an edge so as not to perform denoising on the edge, and performs denoising on a region around the edge, in which a slight stair step effect occurs, and thus a visually-naturally connected edge may be generated.
  • FIG. 5 is a flowchart of a block-based image denoising method, according to an exemplary embodiment.
  • In operation 510, from among certain-sized blocks divided from an input image, similarities between a current block and reference blocks within a search range around the current block are determined. The similarities between the current block and the reference blocks may be calculated by using absolute differences between an average value of the current block and the average values of the reference blocks.
  • In operation 520, weights of the reference blocks with respect to the current block are determined based on the similarities. The weights of the reference blocks may be proportional to exponential function values regarding the similarities between the current block and the reference blocks.
  • Also, in order to reduce differences in boundaries between resultant blocks obtained by denoising the current block and neighboring blocks, the weights of the reference blocks may be changed. According to an exemplary embodiment, the weights of the reference blocks may be adjusted by performing soft thresholding.
  • In operation 530, resultant blocks are generated for every block of the input image, by denoising the current block based on the weights of the reference blocks. The resultant blocks may be generated by denoising the current block based on the weights of the reference blocks within the search range and a weighted average of the reference blocks.
  • In the block-based image denoising method, since denoised resultant blocks are generated by performing image processing on blocks that are not overlappedly processed, block-based denoising may be performed based on block-based similarities. Thus, according to an exemplary embodiment, in comparison to a pixel-based image denoising method, an image that is denoised to be naturally connected to neighboring images may be generated. Also, due to block-based calculation, the speed of calculation may be increased in comparison to the pixel-based image denoising method.
  • Furthermore, as image denoising is performed in consideration of image characteristics between a current block and neighboring blocks, a blocking effect that may occur due to block-based image processing may be prevented.
  • Meanwhile, the exemplary embodiments can be written as computer programs and can be implemented in general-use digital computers that execute the programs using a computer readable recording medium. Examples of the computer readable recording medium include magnetic storage media (e.g., ROM, floppy disks, hard disks, etc.), optical recording media (e.g., CD-ROMs, or DVDs), etc.
  • While the exemplary embodiments have been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by one of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the following claims. The exemplary embodiments should be considered in a descriptive sense only and not for purposes of limitation. Therefore, the scope of the invention is defined not by the detailed description of the invention but by the following claims, and all differences within the scope will be construed as being included in the present invention.

Claims (19)

  1. 1. A block-based image denoising method comprising:
    determining similarities between a current block and reference blocks within a search range around the current block, the current block being from among blocks of an input image;
    determining weights of the reference blocks with respect to the current block based on the similarities; and
    generating a resultant block by denoising the current block based on the weights of the reference blocks.
  2. 2. The method of claim 1, wherein the determining of the weights comprises changing the weights of the reference blocks to reduce differences in boundaries between the resultant block obtained by denoising the current block, and neighboring blocks.
  3. 3. The method of claim 1, wherein the determining of the similarities comprises determining a similarity between the current block and each of neighboring blocks disposed within the search range around the current block.
  4. 4. The method of claim 1, wherein the determining of the similarities comprises determining a sum of absolute differences (SAD) between the current block and the reference blocks.
  5. 5. The method of claim 4, wherein the determining of the weights comprises determining exponential function values with respect to the SAD between the current block and the reference blocks, as the weights of the reference blocks.
  6. 6. The method of claim 2, wherein the changing of the weights comprises changing the weights of the reference blocks by performing soft thresholding, before regularization is performed to calculate a weighted average of the reference blocks with respect to the current block.
  7. 7. The method of claim 1, wherein the generating of the resultant block comprises generating the resultant block with respect to the current block by using the reference blocks and a weighted average of the weights of the reference blocks.
  8. 8. The method of claim 1, wherein denoising is performed on each of color channels of the input image.
  9. 9. A block-based image denoising apparatus comprising:
    a similarity determination unit which determines similarities between a current block and reference blocks within a search range around the current block, the current block being from among blocks of an input image;
    a weight determination unit which determines weights of the reference blocks with respect to the current block based on the similarities; and
    a resultant block generation unit which generates a resultant block by denoising the current block based on the weights of the reference blocks.
  10. 10. The apparatus of claim 9, wherein the weight determination unit changes the weights of the reference blocks to reduce differences in boundaries between the resultant block obtained by denoising the current block, and neighboring blocks.
  11. 11. The apparatus of claim 9, wherein the similarity determination unit determines a similarity between the current block and each of neighboring blocks disposed within the search range around the current block.
  12. 12. The apparatus of claim 9, wherein the similarity determination unit determines a sum of absolute differences (SAD) between the current block and the reference blocks.
  13. 13. The apparatus of claim 12, wherein the weight determination unit determines exponential function values with respect to the SAD between the current block and the reference blocks, as the weights of the reference blocks.
  14. 14. The apparatus of claim 10, wherein the weight determination unit changes the weights of the reference blocks by performing soft thresholding, before regularization is performed to calculate a weighted average of the reference blocks with respect to the current block.
  15. 15. The apparatus of claim 9, wherein the resultant block generation unit generates the resultant block with respect to the current block by using the reference blocks and a weighted average of the weights of the reference blocks.
  16. 16. The apparatus of claim 9, wherein denoising is performed on each of color channels of the input image.
  17. 17. A computer readable recording medium having recorded thereon a computer program for executing the method of claim 1.
  18. 18. The method of claim 6, wherein the soft thresholding comprises:
    comparing one of the weights of the reference blocks to a threshold; and
    reducing the one of the weights of the reference blocks by the threshold if the one of the weights of the reference blocks is greater than the threshold or reducing the one of the weights of the reference blocks to zero if the one of the weights of the reference blocks is not greater than the threshold.
  19. 19. The apparatus of claim 14, wherein in the soft thresholding, one of the weights of the reference blocks is compared to a threshold, and if the one of the weights of the reference blocks is greater than the threshold, reducing the one of the weights of the reference blocks by the threshold, and if the one of the weights of the reference blocks is not greater than the threshold, reducing the one of the weights of the reference blocks to zero.
US12913909 2009-12-16 2010-10-28 Method and apparatus for block-based image denoising Expired - Fee Related US8818126B2 (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
KR10-2009-0125701 2009-12-16
KR20090125701A KR101674078B1 (en) 2009-12-16 2009-12-16 Method and apparatus for block-based image denoising

Publications (2)

Publication Number Publication Date
US20110142368A1 true true US20110142368A1 (en) 2011-06-16
US8818126B2 US8818126B2 (en) 2014-08-26

Family

ID=44142996

Family Applications (1)

Application Number Title Priority Date Filing Date
US12913909 Expired - Fee Related US8818126B2 (en) 2009-12-16 2010-10-28 Method and apparatus for block-based image denoising

Country Status (2)

Country Link
US (1) US8818126B2 (en)
KR (1) KR101674078B1 (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110249909A1 (en) * 2010-04-12 2011-10-13 Novatek Microelectronics Corp. Filter and filtering method for reducing image noise
US20140226905A1 (en) * 2013-02-14 2014-08-14 Canon Kabushiki Kaisha Image processing apparatus, image processing method, and storage medium
US20150093041A1 (en) * 2013-09-30 2015-04-02 Samsung Electronics Co., Ltd. Method of reducing noise in image and image processing apparatus using the same
US9002136B2 (en) 2012-06-29 2015-04-07 Samsung Electronics Co., Ltd. Denoising apparatus, system and method
WO2016048464A1 (en) * 2014-09-23 2016-03-31 Intel Corporation Non-local means image denoising with detail preservation using self-similarity driven blending
US10043252B2 (en) 2016-06-14 2018-08-07 Intel Corporation Adaptive filtering with weight analysis

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3167429A4 (en) 2015-05-15 2017-08-02 SZ DJI Technology Co., Ltd. System and method for supporting image denoising based on neighborhood block dimensionality reduction
KR20170107252A (en) * 2016-03-15 2017-09-25 삼성전자주식회사 A method for removing noise of image data

Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6188799B1 (en) * 1997-02-07 2001-02-13 Matsushita Electric Industrial Co., Ltd. Method and apparatus for removing noise in still and moving pictures
US20060159351A1 (en) * 2005-01-14 2006-07-20 Lg Electronics Inc. Apparatus and method for compensating images in display device
US20080267530A1 (en) * 2007-04-27 2008-10-30 Suk Hwan Lim Generating compound images having increased sharpness and reduced noise
US20080285655A1 (en) * 2006-05-19 2008-11-20 The Hong Kong University Of Science And Technology Decoding with embedded denoising
US20080285868A1 (en) * 2007-05-17 2008-11-20 Barinder Singh Rai Simple Adaptive Wavelet Thresholding
US20090041376A1 (en) * 2007-08-03 2009-02-12 Joan Elizabeth Carletta Method for real-time implementable local tone mapping for high dynamic range images
US7580077B2 (en) * 2005-02-08 2009-08-25 Imagination Technologies Limited Conversion of video data from interlaced to non-interlaced format
US7734115B2 (en) * 2005-12-08 2010-06-08 Industry-Academic Cooperation Foundation, Yonsei University Method for filtering image noise using pattern information
US20100246691A1 (en) * 2009-03-26 2010-09-30 Apple Inc. Restore filter for restoring preprocessed video image
US20100266041A1 (en) * 2007-12-19 2010-10-21 Walter Gish Adaptive motion estimation
US7903179B2 (en) * 2002-06-25 2011-03-08 Panasonic Corporation Motion detection device and noise reduction device using that
US20110063517A1 (en) * 2009-09-15 2011-03-17 Wei Luo Method And System For Utilizing Non-Local Means (NLM) For Separation Of Luma (Y) And Chroma (CBCR) Components
US8023562B2 (en) * 2007-09-07 2011-09-20 Vanguard Software Solutions, Inc. Real-time video coding/decoding
US8086052B2 (en) * 2003-05-20 2011-12-27 Peter Toth Hybrid video compression method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101338612B1 (en) 2007-08-08 2013-12-06 엘지디스플레이 주식회사 Device and method for removing noise in video, liquid crystal display device having the same

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6188799B1 (en) * 1997-02-07 2001-02-13 Matsushita Electric Industrial Co., Ltd. Method and apparatus for removing noise in still and moving pictures
US7903179B2 (en) * 2002-06-25 2011-03-08 Panasonic Corporation Motion detection device and noise reduction device using that
US8086052B2 (en) * 2003-05-20 2011-12-27 Peter Toth Hybrid video compression method
US20060159351A1 (en) * 2005-01-14 2006-07-20 Lg Electronics Inc. Apparatus and method for compensating images in display device
US7580077B2 (en) * 2005-02-08 2009-08-25 Imagination Technologies Limited Conversion of video data from interlaced to non-interlaced format
US7734115B2 (en) * 2005-12-08 2010-06-08 Industry-Academic Cooperation Foundation, Yonsei University Method for filtering image noise using pattern information
US20080285655A1 (en) * 2006-05-19 2008-11-20 The Hong Kong University Of Science And Technology Decoding with embedded denoising
US20080267530A1 (en) * 2007-04-27 2008-10-30 Suk Hwan Lim Generating compound images having increased sharpness and reduced noise
US20080285868A1 (en) * 2007-05-17 2008-11-20 Barinder Singh Rai Simple Adaptive Wavelet Thresholding
US20090041376A1 (en) * 2007-08-03 2009-02-12 Joan Elizabeth Carletta Method for real-time implementable local tone mapping for high dynamic range images
US8023562B2 (en) * 2007-09-07 2011-09-20 Vanguard Software Solutions, Inc. Real-time video coding/decoding
US20100266041A1 (en) * 2007-12-19 2010-10-21 Walter Gish Adaptive motion estimation
US20100246691A1 (en) * 2009-03-26 2010-09-30 Apple Inc. Restore filter for restoring preprocessed video image
US20110063517A1 (en) * 2009-09-15 2011-03-17 Wei Luo Method And System For Utilizing Non-Local Means (NLM) For Separation Of Luma (Y) And Chroma (CBCR) Components

Non-Patent Citations (7)

* Cited by examiner, † Cited by third party
Title
Abstract of IEEE document titled "A Non-Local Algorithm for Image Denoising", authored by Buades et al, published in 2005, 1 page *
Buadeas and Morel, "A Non-Local Algorithm for Image Denoising", 2005, Pages 4-6 *
Coupé, Yger, Prima, Hellier, Kervrann, and Barillot, "An Optimized Blockwise Nonlocal Means Denoising Filter for 3-D Magnetic Resonance Images", April 2008, IEEE Transactions on Medical Imaging, Vol. 27, NO.4, Pages 425-441 *
Cover of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, volume 2, issued 20-25 June 2005, 1 page *
Donoho, "De-Noising by Soft-Thresholding", May 1995, IEEE Transaction on Information Theory, Vol. 41, NO.3, Pages 613-627 *
Lukin, A. "A Multiresolution Approach for Improving Quality of Image Denoising Algorithms," 19 May 2006, IEEE ICASSP 2006, volume 2, 5 pages *
Mahmoudi and Sapiro, "Fast Image and Video Denoising via Nonlocal Means of Similar Neighborhoods", December 2005, IEEE Signal Processing Letters, Col. 12, NO.12, Pages 839-842 *

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110249909A1 (en) * 2010-04-12 2011-10-13 Novatek Microelectronics Corp. Filter and filtering method for reducing image noise
US9002136B2 (en) 2012-06-29 2015-04-07 Samsung Electronics Co., Ltd. Denoising apparatus, system and method
US20140226905A1 (en) * 2013-02-14 2014-08-14 Canon Kabushiki Kaisha Image processing apparatus, image processing method, and storage medium
US9747671B2 (en) 2013-02-14 2017-08-29 Canon Kabushiki Kaisha Image processing apparatus, image processing method, and storage medium
US9384531B2 (en) * 2013-02-14 2016-07-05 Canon Kabushiki Kaisha Image processing apparatus, image processing method, and storage medium for reducing noise in a captured image
US20170316550A1 (en) * 2013-02-14 2017-11-02 Canon Kabushiki Kaisha Image processing apparatus, image processing method, and storage medium
US9330442B2 (en) * 2013-09-30 2016-05-03 Samsung Electronics Co., Ltd. Method of reducing noise in image and image processing apparatus using the same
US20150093041A1 (en) * 2013-09-30 2015-04-02 Samsung Electronics Co., Ltd. Method of reducing noise in image and image processing apparatus using the same
WO2016048464A1 (en) * 2014-09-23 2016-03-31 Intel Corporation Non-local means image denoising with detail preservation using self-similarity driven blending
US9489720B2 (en) 2014-09-23 2016-11-08 Intel Corporation Non-local means image denoising with detail preservation using self-similarity driven blending
US10043252B2 (en) 2016-06-14 2018-08-07 Intel Corporation Adaptive filtering with weight analysis

Also Published As

Publication number Publication date Type
US8818126B2 (en) 2014-08-26 grant
KR101674078B1 (en) 2016-11-08 grant
KR20110068645A (en) 2011-06-22 application

Similar Documents

Publication Publication Date Title
Mahmoudi et al. Fast image and video denoising via nonlocal means of similar neighborhoods
Werlberger et al. Motion estimation with non-local total variation regularization
Liu et al. Automatic estimation and removal of noise from a single image
He et al. Fast matting using large kernel matting laplacian matrices
Buades et al. Neighborhood filters and PDE’s
Yang et al. An efficient TVL1 algorithm for deblurring multichannel images corrupted by impulsive noise
US20080152296A1 (en) Methods and Systems for Processing Film Grain Noise
US20080298704A1 (en) Face and skin sensitive image enhancement
US20120219229A1 (en) Image enhancement apparatus and method
US20110090352A1 (en) Image deblurring using a spatial image prior
US20060291741A1 (en) Image processing apparatus, image processing method, program, and recording medium therefor
US20100245670A1 (en) Systems and methods for adaptive spatio-temporal filtering for image and video upscaling, denoising and sharpening
US8244054B2 (en) Method, apparatus and integrated circuit capable of reducing image ringing noise
Hajiaboli An anisotropic fourth-order diffusion filter for image noise removal
US20110102642A1 (en) Image deblurring using a combined differential image
Guo et al. Adaptive Perona-Malik model based on the variable exponent for image denoising
Sutour et al. Adaptive regularization of the NL-means: Application to image and video denoising
US20100135575A1 (en) Apparatus and method for reducing artifacts in images
US20110075935A1 (en) Method to measure local image similarity based on the l1 distance measure
US20080118179A1 (en) Method of and apparatus for eliminating image noise
US6323855B1 (en) Sharpening edge features in digital image providing high frequency edge enhancement
Kawai et al. Image inpainting considering brightness change and spatial locality of textures and its evaluation
Mignotte A segmentation-based regularization term for image deconvolution
US6959117B2 (en) Method and apparatus for deblurring and re-blurring image segments
US20100166300A1 (en) Method of generating motion vectors of images of a video sequence

Legal Events

Date Code Title Description
AS Assignment

Owner name: SAMSUNG ELECTRONICS CO., LTD., KOREA, REPUBLIC OF

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:SHIN, GUN-SHIK;KIM, DONG-HYUN;LEE, JAE-HUN;AND OTHERS;SIGNING DATES FROM 20100726 TO 20100729;REEL/FRAME:025208/0808

FEPP

Free format text: MAINTENANCE FEE REMINDER MAILED (ORIGINAL EVENT CODE: REM.)

LAPS Lapse for failure to pay maintenance fees

Free format text: PATENT EXPIRED FOR FAILURE TO PAY MAINTENANCE FEES (ORIGINAL EVENT CODE: EXP.); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY